Skip to main content

Posts

Showing posts from July, 2020

On 48 hours testing

Bill Gates says most coronavirus tests are a 'complete waste' because the results come back too slow
From article:
"The simplest thing, which has to do with such insanity, is you should not reimburse somebody for getting a test that it takes more than 48 hours to get the result back," Gates said on Tuesday. "That test is a complete waste. And to all these numbers about how much we test, the majority is just complete waste," he added, calling it "insane" to have to pay for test results that could take more than three days and up to a full week.


My opinion:
I can't say I completely agree. First, if you test yourself, be responsible and until you have results do not meet with anyone. It's all about self responsibility. Also, even if your test comes negative, you want to meet with as few people as possible during these times.
And most importantly, results of these tests might be useful for "aftermath" of pandemic, when trying to better und…

When you realize you need only 64 random convolutions

when searching for 8 by 8 convolutions, as with 64 free parameters for every one of them you can create output of any convolution by linear combination of outputs from 64 random convolutions (which are not trained). 
Config:
convolution,1,28,28,64,8,8,0.5,0 convolution,1,64,441,200,64,1,0.5,-0.001 max,200,21,21,7,7 matrix,1800,130,0.5,-0.001 sigmoid,130 dropout,130,0.5 matrix,130,10,0.5,-0.001 softmax,10
First convolution is not trained, and you need only 64 of them (8x8 parameters = 64). Then with next convolution you find convolutions which you need by combining outputs of previous convolutions.
Results (above 99%):

Cool visualizations of objects in low earth orbit

When you don't train parameters of convolution in neural network

So, I tried very simple experiment, where parameters of convolution are initialized randomly and not trained at all. The results are better than what I would expect on MNIST dataset.
convolution,1,28,28,400,8,8,0.5,0 max,400,21,21,7,7 matrix,3600,130,0.5,-0.001 sigmoid,130 dropout,130,0.5 matrix,130,10,0.5,-0.001 softmax,10
0 at the end for convolution means 0 step size - do not train. Only parameters of other layers are trained. Trained with WideOpenThoughts.
Results (above 99% accuracy for MNIST):

When you do your own haircut and you have only 3mm comb at home

I can feel the wind

When you lose interest in Neuralink's research

Last couple days, there was a lot of articles in news about Neuralink and listening to music with implants. I personally have no interest to get implant and make a private company easy access to my brain and possibility that they might try to control me. If some people think AI is dangerous, I find brain implants way more dangerous. Besides dangerous, I find them also unnecessary. Also, AI is here only to help, not to replace us or make humans not valued...

When you learn something new

In past, I wrote 'Altitude, not attitude'. Today, I read that some people actually use this term in publications 'TAXONOMY AND CLASSIFICATION SCHEME FOR ARTIFICIAL SPACE OBJECTS': "The attitude state is the characteristic how the attitude of the object is evolving over time, is it actively stabilized, naturally stable, spinning moderately around one axis, spinning in irregular manner."
Interesting word choice. I would probably choose different word. Ha.
The main definition from Google: "a settled way of thinking or feeling about someone or something, typically one that is reflected in a person's behavior." is more connected with human behavior rather than non-living object's properties.

Trippy, sound & safe

Together mode

When you find interesting article, but ...

ELON MUSK'S WORLDVIEW ON A.I. EXPLAINED IN 6 MINUTES
From article:
"Elon Musk believes the universe is simulatable. It's just a matter of having the right inputs."
"Musk's comments are fascinating, in part because it highlights a particular worldview. Rather than scientific equations aiming to describe the world as observed, Musk suggests that there are knowable "true" equations that govern the universe waiting to be discovered. It's just a matter of finding the right ones."
My comment:
Even if you would find the "true equations", and we might never find them, we might still not be able to solve them. Knowing equation doesn't mean we can solve it. Often we can only approximate solution.
Further from article:
Musk is not exactly alone in this: Robbert Dijkgraaf, director of the Institute for Advanced Study, explained in Quanta Magazine how Albert Einstein believed that there was a single, unique way to construct a functional universe…

Running Optirol (Windows program) on Linux using Wine

Few months back I tried running Windows program on Linux and it was closer to fail than to succeed feeling. So this time, I tried Wine. Installation of Wine was simple through "Ubuntu Software". On first try, I got error about missing Mono project. After googling problem, I found nice solution.
sudo apt-get install winetricks winetricks dotnet45 After doing that, with many error messages, dot Net was installed so I executed:
wine Optirol.exe To my shock, it actually does work:





The only thing that does not work properly seems to be OpenGL:


Little bummer, because seeing 3D graphs is pretty, but the rest seems to work just fine.

How to turn off Microsoft OneDrive

If you are not using OneDrive, it might be annoying for you as it is for me when you start Windows to see OneDrive - log into your account (or in my case we have troubles logging into your account). To switch it off, just head to "Startup Apps" and you can disable it, as it seems to be enabled by default.

Loop cable for transportation between Moon and Earth

Imagine having an elevator which just hangs in the air in between of Moon and Earth (not attached to either of them). One side pulled by gravitation of the Moon, one side pulled by gravitation of the Earth (gravitational forces of same size, but opposite direction: F1 = -F2). Just a loop cable and motor hanging on the Moon's side (also just hanging in the air with cable). Sure, there is a lot of details that would need to be worked out. But just imagine every night seeing an elevator hanging in the air transporting goods between Moon and Earth. Would that not be trippy?
Space Elevator

On foldable desings of phones

Apple filed a patent for foldable phone (news, patent). Yes, I am super excited about foldable phones, but here is why I don't like this design as much as I like design from Huawei. It's all about durability. If you fold the phone with screen inside, at the point of fold, there will be more pressure on screen at the place due to bigger folding angles and smaller length of the fold, while if you fold with display outside, the pressure is distributed over longer length and the angles are smaller. Hence, design will allow it for more folds and hence it should be more durable. The only advantage of folding the screen inside is protection of it, but we can always have keys in other pocket than our phones, can't we? Winner: Huawei design. I understand that Apple might have come late to game and the patent for having the screen folded outside is gone... So now the only question is if Apple can actually design foldable screen, which folds inside and can be durable or give up and w…

C# execution vs Python execution (performance)

So few days ago I wrote about how python is slower than C#.So I created simple script for python using strings and dictionaries:from datetime import datetime
import random
dic = {}
letter = 'qwertyuiopasdfghjklz'
for i in range(10000001):
    if i%1000000==0:
        now = datetime.now()
        print(str(now))

    s = ""
    for j in range(50):
        s += letter[random.randrange(20)]
    if s not in dic:
        dic[s] = True
And same thing in C#:using System;
using System.Collections.Generic;
namespace ConsoleApp1
{
    class Program
    {
        static void Main(string[] args)
        {
            Dictionary<string, bool> dic = new Dictionary<string, bool>();
            Random r = new Random();
            string letter = "qwertyuiopasdfghjklz";
            for (int i=0;i<10000001;i++)
            {
                if (i % 1000000 == 0)
                {
                    Console.WriteLine(DateTime.Now.ToString());
                }
                string s = &q…

Now that's a cool idea

When you get really excited II

According to ESA’s Solar Orbiter Project Scientist Daniel Müller, the images, to be released in mid-July, will be the closest images of the Sun ever captured.
“Our ultraviolet imaging telescopes have the same spatial resolution as those of NASA’s Solar Dynamic Observatory (SDO), which takes high-resolution images of the Sun from an orbit close to Earth. Because we are currently at half the distance to the Sun, our images have twice SDO’s resolution during this perihelion,” says Daniel.

Mid July is close by.

Windows vs Ubuntu performance on the same laptop

Last few days, I switched to Windows again (yeah it's the C# addiction - but I believe that's healthy addiction so it's all good in my book) and there is one thing I did notice when comparing Windows and Ubuntu. I have to say it was easy to spot it and that is responsiveness of UI and Windows wins this category. Another difference, and this one was more noticeable, I noticed is opening csv files in Microsoft Office vs LibreOffice and Microsoft Office not only opens up quite faster, but displays the graphs faster as well and manipulating graphs was more responsive. Last but not least and again quite a difference, execution of C# .Net prototype vs Python prototype - C# .Net won big time. Seems like Microsoft does one thing right and that is caring about performance of their software.